Spremljaj
Amartya Sanyal
Naslov
Navedeno
Navedeno
Leto
Calibrating Deep Neural Networks using Focal Loss
J Mukhoti, V Kulharia, A Sanyal, S Golodetz, PHS Torr, PK Dokania
Advances in Neural Information Processing Systems (NeurIPS), December 2020, 2020
4732020
TAPAS: Tricks to Accelerate (encrypted) Prediction As a Service
A Sanyal, MJ Kusner, A Gascón, V Kanade
International Conference on Machine Learning 80, 4497--4506, 2018
1652018
Progressive skeletonization: Trimming more fat from a network at initialization
P de Jorge, A Sanyal, HS Behl, PHS Torr, G Rogez, PK Dokania
International Conference on Learning Representations, ICLR 2021, 2021
1022021
How benign is benign overfitting?
A Sanyal, PK Dokania, V Kanade, PHS Torr
International Conference on Learning Representations, (Spotlight Paper) ICLR …, 2021
642021
Make Some Noise: Reliable and Efficient Single-Step Adversarial Training
P de Jorge, A Bibi, R Volpi, A Sanyal, PHS Torr, G Rogez, PK Dokania
Advances in Neural Information Processing Systems (NeurlPS), 2022
58*2022
Stable Rank Normalization for Improved Generalization in Neural Networks and GANs
A Sanyal, PHS Torr, PK Dokania
International Conference on Learning Representations, (Spotlight Paper) ICLR …, 2019
532019
Towards Adversarial Evaluations for Inexact Machine Unlearning
S Goel, A Prabhu, A Sanyal, SN Lim, P Torr, P Kumaraguru
arXiv preprint arXiv:2201.06640, 2022
41*2022
Optimizing non-decomposable measures with deep networks
A Sanyal, P Kumar, P Kar, S Chawla, F Sebastiani
Machine Learning 107, 1597-1620, 2018
392018
How robust is unsupervised representation learning to distribution shift?
Y Shi, I Daunhawer, JE Vogt, P Torr, A Sanyal
International Conference on Learning Representations (ICLR), 2023
34*2023
Robustness via Deep Low-Rank Representations
A Sanyal, V Kanade, PHS Torr, PK Dokania
arXiv preprint arXiv:1804.07090, 2018
34*2018
How unfair is private learning ?
A Sanyal, Y Hu, F Yang
Conference on Uncertainty in Artificial Intelligence (Oral Paper) UAI, 2022
282022
Corrective machine unlearning
S Goel, A Prabhu, P Torr, P Kumaraguru, A Sanyal
arXiv preprint arXiv:2402.14015, 2024
142024
Multiscale sequence modeling with a learned dictionary
B van Merriënboer, A Sanyal, H Larochelle, Y Bengio
ICML 2017 Workshop on Machine Learning in Speech and Language Processing, 2017
122017
PILLAR: How to make semi-private learning more effective
F Pinto, Y Hu, F Yang, A Sanyal
Conference on Secure and Trustworthy Machine Learning (SatML) 2024, 2023
112023
A law of adversarial risk, interpolation, and label noise
D Paleka, A Sanyal
International Conference on Learning Representations (ICLR) 2023, 2023
92023
Catastrophic overfitting can be induced with discriminative non-robust features
G Ortiz-Jimenez, P de Jorge, A Sanyal, A Bibi, PK Dokania, P Frossard, ...
Transactions on Machine Learning Research, 2023
6*2023
Certified private data release for sparse Lipschitz functions
K Donhauser, J Lokna, A Sanyal, M Boedihardjo, R Hönig, F Yang
International Conference on Artificial Intelligence and Statistics, 1396-1404, 2024
5*2024
Can semi-supervised learning use all the data effectively? A lower bound perspective
A Tifrea, G Yüce, A Sanyal, F Yang
Advances in Neural Information Processing Systems 36, 2024
42024
What makes and breaks safety fine-tuning? mechanistic study
S Jain, ES Lubana, K Oksuz, T Joy, PHS Torr, A Sanyal, PK Dokania
arXiv preprint arXiv:2407.10264, 2024
22024
How robust accuracy suffers from certified training with convex relaxations
P De Bartolomeis, J Clarysse, A Sanyal, F Yang
arXiv preprint arXiv:2306.06995, 2023
22023
Sistem trenutno ne more izvesti postopka. Poskusite znova pozneje.
Članki 1–20